Monday, March 28, 2005

New Mayo Clinic Study Reveals Problems with Lung Cancer Screening

An article in the April issue of Radiology reports the results of a study of the use of helical CT scanning as an approach for the early detection of lung cancer among high risk smokers (Swensen SJ, Jett JR, Hartman TE, et al. CT screening for lung cancer: five-year prospective experience. Radiology 2005; 235:259-265). A group of 1,520 adults ages 50+ was followed for four years, receiving a baseline chest CT scan and four annual follow-up CT scans. The purpose of the study was to determine whether this approach would save lives by detecting lung cancers earlier than they would otherwise be detected, allowing for earlier and more effective treatment.

The study found that CT scan screening did not significantly reduce lung cancer deaths. Nor did it result in a shift to the detection of more very early cancers. But this screening did create problems for patients because of a very high false positive rate. False positive tests mean the finding of a lung nodule on the CT scan that turns out not to be cancerous. About three-fourths of the patients had a lung nodule detected, and of these, only 4% turned out to have cancer. Thus, well over 1,000 cancer-free patients were told they had a lung nodule suspicious for cancer and had to have further diagnostic procedures, sometimes involving surgical or other procedures (i.e., biopsy).

The Rest of the Story

While chest CT scan screening of current and ex-smokers at high risk for lung cancer turned out not to have any benefit to them, it did cause significant morbidity and mental distress. Being told that one has a nodule that could be cancerous tends to create substantial anxiety. For many of these patients, they had to live with this unnecessary anxiety for six months or even a year before they were able to be told that there was nothing to worry about. For others, they had to undergo further diagnostic tests, sometimes surgery.

Even among patients with new lung nodules greater than 4mm in size detected (nodules that were not present on the initial CT scan), the false positive rate was 92%. This means that the overwhelming majority of patients with newly diagnosed lung nodules (92%) did not have lung cancer.

Because of the anxiety produced by being told that one has a potentially cancerous lung nodule as well as the significant morbidity (and even mortality) associated with some of the interventions used to definitively diagnose these nodules, this study reveals that the CT screening approach resulted in more harm than good for these patients.

In order for a screening test to be effective for a relatively uncommon disease (and lung cancer is still uncommon, even in high risk smokers), the false positive rate must be extremely low. In other words, the test must not detect a large number of nodules that turn out to be benign. Otherwise, these patients will suffer a great deal of unnecessary anxiety and possibly morbidity and mortality. There would certainly have to be strong evidence of a significant mortality benefit in the population to recommend the use of such a screening test. In this case, helical CT scanning simply does not meet these criteria.

In more general terms, the results of this study underscore the pointlessness of devoting substantial resources to lung cancer screening. We should be focusing our resources in the area where we know we can have an effect: prevention of lung cancer through interventions to prevent smoking initiation and encourage smoking cessation.

A central tenet of medicine is "first, to do no harm." Screening for lung cancer, even in high risk individuals, appears to violate this basic principle.


Fred Grannis MD said...


Respectfully, you are all wrong on the state of the art in lung cancer screening.

The current U.S. system of dealing with lung cancer is to do NOTHING until the patient walks into a doctors office with symptoms of lung cancer. At that point in time, 40% have stage IV disease and another 35% have stage III disease. Current five year survival in stage IV is 0% and approximately 5% in stage III. This strategy is completely bankrupt and only 14% of LC patients currently reach 5 yrs. This figure has increased only marginally over the past 30 yrs. and WILL NOT improve further without implementation of screening.

ALthough I am a graduate of Mayo and a friend of Jim Jett, this is a weak study. Mayo did not have a protocol to manage the screening PROCESS effectively. Screening is not a TEST, it is a PROCESS. The International Early Lung Cancer Action Program, in which I participate as a PI, has taken a different approach and used a protocol that is updated every six months based upon accumulating experience, and is available in full text or diagram for your readers on the web at

Using the protocol to dispell anxiety and avoid unnecessary testing and surgery, the I-ELCAP consortium has identified more than 400 lung cancers, with screening of 27,000+ individuals. There is a marked shift in diagnosis to early stage, with more than 80% detected in stage I. Survival data will be published in the near future, and will show a striking improvment in survival in screen detected LC.

As you know, I am a strong advocate of tobacco control policy and measures to reduce death and disability from disease caused by tobacco products, but it is important for tobacco control experts to understand, that there are more than 90 million Americans at risk of LC. Reduction of smoking in this population will reduce LC deaths, but without implementation of effective screening, 160,000 of these individuals will continue to suffer and die from lung cancer each year, far into the forseeable future. The mortality worldwide will be staggering.

We have the technology and the knowledge to screen for lung cancer NOW. For advocates of screening and tobacco control experts to squabble over the skimpy revenues available is counterproductive to both groups, but most of all works against saving the lives of those who, without screening, are currently doomed to lung cancer deaths.

Fred Grannis MD
City of Hope
Duarte CA

Michael Siegel said...

Thanks so much for these comments. I appreciate your taking the time to share your perspective on this issue. Since the more recent ELCAP results have not been published yet, I obviously could not take them into account in this commentary. However, as you know, it is always important to continually re-evaluate the state of evidence, so as new results are published, I will certainly re-evaluate and comment at that time (and will change my opinion if the results suggest that is appropriate).

The one thing I would point out is that the evaluation of a screening test is not based only on whether it is effective in detecting early and treatable disease, but also on whether it is specific enough to avoid a large rate of false positive results. In the case of the first ELCAP paper, the positive predictive value was only 12%, meaning that 88% of patients who were told that they had a lung nodule turned out not to have lung cancer. Because of the significant anxiety associated with being told that one has a lung nodule and having to wait a year or perhaps longer to get a definitive diagnosis, I still question whether this approach is reasonable when applied on a population level.

Anonymous said...


I disagree that "anxiety" among the 92% with a false positive CT screening tests presents a greater public health risk than the combined public health benefits from the early detection and treatment of lung cancer among the 8% with true positives, and the preventive public health benefits from people quitting smoking due to getting the CT scan, and due to getting a false positive.

As is the case with HIV tests, I suspect that nearly all of those who get the CT scan confront different levels of anxiety just by taking the test.

And while a false positive probably generates even greater anxiety (than by just taking the CT scan), that anxiety can be reduced by informing the people that there is a 92% false positive rate.

But that anxiety can encourage many smokers to quit, or to at least try to quit. My point is that anxiety can be an effective motivator, and shouldn't necessarily be viewed as a bad thing.

But I don't think that or other CT scan studies even measure the smoking cessation benefits among those scanned, or of those with a false positive.

Bill Godshall

Michael Siegel said...

You make a very good point, and I agree with you that the public health risks associated with anxiety among the 92% with a false positive test do not exceed the public health risks of failing to detect the 8% with lung cancer early enough for possible definitive treatment. But there is also a question about the equity of distribution of benefits and costs of a screening test when it is applied on a population level. I'd be curious to hear your thoughts on that issue. For example, would you support the use of CT screening for all populations at increased risk of cancer (such as those with high fat diets or low antioxidant intake). The routine use of whole body CT could certainly detect a massive number of cancers early, but it would come at the expense of huge numbers of healthy people with false positive tests who would be bearing the burden for those whose lives would be saved. It's certainly a difficult trade-off to have to make.